GRASPEL_matlab.zip contains the GRASPEL code implemented in MATLAB. Experimental results in the paper are based on this version of code. Run runGRASPEL.m for the GRASPEL algorithm example on the USPS data set.

Implementation of GRASPEL algorithm into Julia Preprocessing and loading set up for included USPS dataset as matfile (USPS.mat)

include GRASPEL.JL for the primary function which returns G, a, varRatio[1:numIter], costVec[1:numIter], numIter

FiedlerConstruct(fea, samplePerc, edge_iter, sig, maxIterations, k, tol, num_node, dim)

fea is the vector of n-dimensional inputs
samplePerc is the percentage of sorted nodes as candidate edges
edge_iter is the edge budget per iteration (percentage of nodes)
sig is the feature variance
maxIterations is the max number of GRASPEL iterations to perform
k is the positive integer size of the original k-nearest-neighbor graph
tol is the embedding distortion tolerance (should be > 1)
num_node is the length of the input vector
dim is the length of an element of the input vector

This function returns G, varRatio[1:numIter], costVec[1:numIter], numIter

G is a SimpleWeightedGraphs object of the resulting graph
varRatio is a history of the varRatio through the iterations
costVec is a history of the costVec through the iterations
numIter is the number of completed iterations

The example RUN_USPS.jl preprocesses and runs the data in the USPS.mat file, which is the USPS dataset of handwritten digits.

Lines 7-12 in the RUN_USPS dataset set the parameters in the FiedlerConstruct function

1. GRASPEL.jl, RUN_USPS.jl, and USPS.mat must be in the same directory

2. The following packages must be installed in the Julia REPL

   --MAT https://github.com/JuliaIO/MAT.jl
   --HNSW https://github.com/JuliaNeighbors/HNSW.jl
   --Distances https://github.com/JuliaStats/Distances.jl
   --MatrixMarket https://github.com/JuliaSparse/MatrixMarket.jl
   --Graphs https://github.com/JuliaGraphs/Graphs.jl
   --SimpleWeightedGraphs https://github.com/JuliaGraphs/SimpleWeightedGraphs.jl
   --Arpack https://github.com/JuliaLinearAlgebra/Arpack.jl
   --Clustering https://github.com/JuliaStats/Clustering.jl

   command:

    add MAT, HNSW, Distances, MatrixMarket, Graphs, SimpleWeightedGraphs, Arpack, Clustering 
   

   after opening the REPL and pressing "]" to enter pkg editing

3. Navigate to the directory in terminal and run RUN_USPS.jl using the command: "julia RUN_USPS.jl"

4. The function should create two matrixmarket (.mtx) files named "Julia_A.mtx" and "Julia_L.mtx", for the Laplacian and Adjacency matrices of the resulting graph G.